Reconstruction of fluorescence molecular tomography with a cosinoidal level set method.

Reconstruction of fluorescence molecular tomography with a cosinoidal level set method.
复制标题

余弦水平集法重建荧光分子断层扫描

DOI:
10.1186/s12938-017-0377-0
复制
发表时间:
2017-06-27
影响因子:
3.9
通讯作者:
Zhu S
Zhu S
中科院分区:
工程技术3区
文献类型:
--
作者:
Zhang X;Cao X;Zhu S

文献摘要

相似文献

背景在荧光分子断层成像(FMT)中,基于形状的重建方法比基于图像的重建方法能够获得更高的图像清晰度。然而,隐式形状方法由于采用了基于梯度的优化方法,收敛速度慢,计算结果不稳定。方法提出了一种基于余弦水平集方法的基于形状的FMT重建方案。将经典隐形方法中的Heaviside函数用余弦函数代替,再用Levenberg-MarQuardt方法代替基于梯度的方法进行重建。结果通过数值模拟和模型实验验证了该方法的有效性。与隐形方法和基于图像的重建方法相比,该方法对噪声比和皮尔逊相关性具有更高的对比度。结论与基于图像的重建方法相比,该方法性能更稳定,收敛速度更快,图像清晰度更高。
BackgroundImplicit shape-based reconstruction method in fluorescence molecular tomography (FMT) is capable of achieving higher image clarity than image-based reconstruction method. However, the implicit shape method suffers from a low convergence speed and performs unstably due to the utilization of gradient-based optimization methods. Moreover, the implicit shape method requiresprioriinformation about the number of targets.MethodsA shape-based reconstruction scheme of FMT with a cosinoidal level set method is proposed in this paper. The Heaviside function in the classical implicit shape method is replaced with a cosine function, and then the reconstruction can be accomplished with the Levenberg–Marquardt method rather than gradient-based methods. As a result, theprioriinformation about the number of targets is not required anymore and the choice of step length is avoided.ResultsNumerical simulations and phantom experiments were carried out to validate the proposed method. Results of the proposed method show higher contrast to noise ratios and Pearson correlations than the implicit shape method and image-based reconstruction method. Moreover, the number of iterations required in the proposed method is much less than the implicit shape method.ConclusionsThe proposed method performs more stably, provides a faster convergence speed than the implicit shape method, and achieves higher image clarity than the image-based reconstruction method.